Reinforcement Learning Engineer, Whole Body Control

What the role actually is

Figure is hiring a reinforcement learning engineer to develop, train, deploy, and evaluate algorithms for whole-body control on its humanoid robot. The role emphasizes sim-to-real gaps, policy metrics, and robust learned control.

This is a hands-on robot learning role for someone who wants learned policies to move from training environments onto real embodied hardware.

What you would work on

• Train and deploy reinforcement learning algorithms for whole-body control
• Choose observations, actions, and model approaches for humanoid movement
• Identify sim-to-real gaps and improve policy robustness
• Define metrics for evaluating learned control performance

What they are asking for

• Strong background in dynamics, controls, or legged robotics
• Experience with RL algorithms for robotics such as PPO or SAC
• Familiarity with reward shaping, curriculum learning, and domain randomization
• Ability to lead complex controls projects and mentor other engineers

Why this one is worth a look

The user experience of a humanoid starts with motion: balance, recovery, hesitation, and how the robot moves near people. This role is a technical one, but its impact shows up directly in whether the robot feels capable or unnerving.